improvement(search): name Slack authors, read only when needed, and list by date without a bound - #8286
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
There was a problem hiding this comment.
All reported issues were addressed across 13 files
Reply with feedback, questions, or to request a fix.
Fix all with cubic | Re-trigger cubic
|
…ist by date without a bound
…authors by user ID
e47d22b to
86c234c
Compare
|
@cubic-dev-ai review this PR |
@waleedlatif1 I have started the AI code review. It will take a few minutes to complete. |
Summary
conversations.repliesidentifies authors only by user ID, so reads labeled every message "Slack member" and the Assistant spent extra reads working out who said what. Reads now name authors from an embedded profile, elseusers.info(Tier 4,users:read), at most 10 lookups per read and in parallel with the permalink call. A failed lookup keeps the generic label instead of failing the read; aborts still propagate. Bot messages use their documentedusernamecontext_messagescarry their ownauthor_name/user_id; results with surrounding messages now label each line with its author. No extra requests: search still never waits on directory lookups. A lone match is unchangedsortBynewest/oldest with no terms and no dates was rejected, costing a model round trip before it retried withendDate= now. The server now adds that bound itself, the same default Slack'sconversations.history/conversations.repliesapply to an omittedlatest. The request is then identical to the one the model sends on retry and is checked like any other bound. Requests with terms or dates are untouched; relevance-sorted empty requests are still rejectedType of Change
Testing
Live search, contract, MCP and mothership suites pass (5,307 tests); new tests fail on the old code. Lint, 49 audits, block registry, docs manifest and type-check clean
Checklist